Prince Raj

Software Engineer

• Designed and ran structured evaluation workflows for coding LLMs by analysing real GitHub PRs, extracting task metadata (language, category, difficulty, critical files) and validating solutions against golden reference patches.
• Wrote 5 to 10 task-specific binary rubrics per task (40 to 50% correctness-weighted) to grade outputs across correctness, code quality, behavioural reasoning and summary accuracy.
• Ran side-by-side (SxS) evaluations of STEM and coding responses on a weighted framework (instruction following, truthfulness, correctness, writing style, verbosity) to determine model preference.
• Benchmarked internal model outputs against GPT-5.2 and Gemini-3 Pro, produced evidence-based justifications and delivered structured JSON reports used in post-training analysis.

Full Stack Developer

• Built an AI / LLM Evaluation & Feedback Platform used internally to benchmark and compare outputs from multiple large language models.
• Built the benchmarking infrastructure, supporting both static dataset runs and real-time model evaluations through provider APIs.
• Designed multi-turn conversation evaluation workflows capturing per-turn scores and overall conversation quality.
• Built an end-to-end project configuration system controlling task generation, evaluation rules, comparison modes and output formats; delivered clean, validated JSONL for research and analytics.
• Implemented quality-control workflows: mandatory rationales, automated validation checks and manual QC review queues.

Sr. Software Engineer · May 2025 – Jul 2025

• Spearheaded an internal design-to-dev pipeline on a custom Canva workspace, streamlining design handoff and cutting UI inconsistencies by 70%.
• Refactored and secured critical API endpoints for performance, RESTful alignment and resilience.
• Drove zero-downtime backend deployments on AWS with monitoring, autoscaling and rollback, reaching 99.9% uptime.
• Mentored junior developers and ran code reviews focused on scalability and maintainability; contributed to architecture decisions.

Software Engineer · Nov 2024 – Apr 2025

• Built and shipped the official company website with 3D model integration and Framer Motion animations.
• Moved data fetching from REST to GraphQL for a more scalable, structured architecture.
• Connected the Request Service module with Zoho Work Channel for automated lead capture, cutting cross-team response time by 60%.
• Refactored the codebase to remove redundant API calls, cutting initial page load time by 40%; deployed on scalable cloud infra with CI/CD.

SDE I · Mar 2024 – Oct 2024

• Built an AI-powered chat application from the ground up, improving frontend efficiency by 60% and backend performance by 30%.
• Cut AI response latency from 1.86s to 1.2s (35% faster) through API design, caching and performance tuning.
• Led an AWS cost-saving initiative: identified unused resources and set S3 expiration policies, cutting over $50,000 a year.
• Developed and integrated APIs with secure authentication; shipped features on the MERN stack 40% faster.

Sr. Frontend Engineer · Sep 2023 – Feb 2024

• Built modern, responsive React UI components that lifted user engagement metrics by 25%.
• Owned frontend architecture decisions for scalable, maintainable, performance-optimised UI systems.
• Reviewed code and enforced quality, performance and security standards.

Full Stack Developer

• Designed and integrated RESTful APIs, reducing system overload and improving frontend-backend communication by 30%.
• Built real-time communication features on WebSocket for live chat, cutting latency issues by 35%.
• Led a performance overhaul: smoother tab switching, minimal downtime, faster navigation; reduced technical debt by 30% through reviews.

Full Stack Developer

• Optimised payment gateway integration, cutting transaction failures by 25%.
• Built and integrated mobile applications with smooth cross-device performance, reducing user complaints by 60%.
• Designed scalable cloud architecture on Microsoft Azure for AI workloads, improving throughput by 10% while cutting infra cost via autoscaling.